OPEN LINK ↗
// 1h agoNEWS
Richard Sutton leaves Keen for Oak Lab
Reinforcement learning pioneer Richard Sutton is departing John Carmack's Keen Technologies to co-found Oak Lab, a new research venture focused on reinforcement learning and lifelong continual learning. The startup aims to build superintelligence from an agent's first-person experience, distinguishing its approach from mainstream transformer-based AGI efforts.
// ANALYSIS
Sutton's launch of Oak Lab is a purist's rebellion against the transformer-dominated landscape, betting that true AGI requires agent-centric, first-person reinforcement learning rather than passive pre-training.
- –Sutton's departure suggests Keen Technologies and Oak Lab may have diverged on the best path to AGI, with Sutton preferring a purely RL-driven approach.
- –Oak Lab will likely serve as the main vehicle to develop the OaK architecture, focusing on agents that create state/time abstractions and plan with world models.
- –While mainstream AGI efforts focus on scaling transformers, this venture could validate or isolate the lifelong reinforcement learning paradigm.
// TAGS
reinforcement-learningrichard-suttonoak-labkeen-technologiesjohn-carmackagistartups
DISCOVERED
1h ago
2026-07-13
PUBLISHED
1h ago
2026-07-13
RELEVANCE
8/ 10
AUTHOR
mark_k